From bb458c644a59dbba3a1fe59b27106c5e68e1c4bd Mon Sep 17 00:00:00 2001 From: Al Viro Date: Sat, 13 Jul 2013 13:26:37 +0400 Subject: Safer ABI for O_TMPFILE [suggested by Rasmus Villemoes] make O_DIRECTORY | O_RDWR part of O_TMPFILE; that will fail on old kernels in a lot more cases than what I came up with. And make sure O_CREAT doesn't get there... Signed-off-by: Al Viro --- include/uapi/asm-generic/fcntl.h | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) (limited to 'include') diff --git a/include/uapi/asm-generic/fcntl.h b/include/uapi/asm-generic/fcntl.h index 06632beaa6d5..05ac354e124d 100644 --- a/include/uapi/asm-generic/fcntl.h +++ b/include/uapi/asm-generic/fcntl.h @@ -84,10 +84,14 @@ #define O_PATH 010000000 #endif -#ifndef O_TMPFILE -#define O_TMPFILE 020000000 +#ifndef __O_TMPFILE +#define __O_TMPFILE 020000000 #endif +/* a horrid kludge trying to make sure that this will fail on old kernels */ +#define O_TMPFILE (__O_TMPFILE | O_DIRECTORY | O_RDWR) +#define O_TMPFILE_MASK (__O_TMPFILE | O_DIRECTORY | O_CREAT | O_ACCMODE) + #ifndef O_NDELAY #define O_NDELAY O_NONBLOCK #endif -- cgit v1.2.3-58-ga151 From 4f5e65a1cc90bbb15b9f6cdc362922af1bcc155a Mon Sep 17 00:00:00 2001 From: Oleg Nesterov Date: Mon, 8 Jul 2013 14:24:16 -0700 Subject: fput: turn "list_head delayed_fput_list" into llist_head fput() and delayed_fput() can use llist and avoid the locking. This is unlikely path, it is not that this change can improve the performance, but this way the code looks simpler. Signed-off-by: Oleg Nesterov Suggested-by: Andrew Morton Cc: Al Viro Cc: Andrey Vagin Cc: "Eric W. Biederman" Cc: David Howells Cc: Huang Ying Cc: Peter Zijlstra Signed-off-by: Andrew Morton Signed-off-by: Al Viro --- fs/file_table.c | 25 ++++++++++--------------- include/linux/fs.h | 2 ++ 2 files changed, 12 insertions(+), 15 deletions(-) (limited to 'include') diff --git a/fs/file_table.c b/fs/file_table.c index b9a77ad08b4d..b44e4c559786 100644 --- a/fs/file_table.c +++ b/fs/file_table.c @@ -265,18 +265,15 @@ static void __fput(struct file *file) mntput(mnt); } -static DEFINE_SPINLOCK(delayed_fput_lock); -static LIST_HEAD(delayed_fput_list); +static LLIST_HEAD(delayed_fput_list); static void delayed_fput(struct work_struct *unused) { - LIST_HEAD(head); - spin_lock_irq(&delayed_fput_lock); - list_splice_init(&delayed_fput_list, &head); - spin_unlock_irq(&delayed_fput_lock); - while (!list_empty(&head)) { - struct file *f = list_first_entry(&head, struct file, f_u.fu_list); - list_del_init(&f->f_u.fu_list); - __fput(f); + struct llist_node *node = llist_del_all(&delayed_fput_list); + struct llist_node *next; + + for (; node; node = next) { + next = llist_next(node); + __fput(llist_entry(node, struct file, f_u.fu_llist)); } } @@ -306,7 +303,6 @@ void fput(struct file *file) { if (atomic_long_dec_and_test(&file->f_count)) { struct task_struct *task = current; - unsigned long flags; file_sb_list_del(file); if (likely(!in_interrupt() && !(task->flags & PF_KTHREAD))) { @@ -320,10 +316,9 @@ void fput(struct file *file) * fput to avoid leaking *file. */ } - spin_lock_irqsave(&delayed_fput_lock, flags); - list_add(&file->f_u.fu_list, &delayed_fput_list); - schedule_work(&delayed_fput_work); - spin_unlock_irqrestore(&delayed_fput_lock, flags); + + if (llist_add(&file->f_u.fu_llist, &delayed_fput_list)) + schedule_work(&delayed_fput_work); } } diff --git a/include/linux/fs.h b/include/linux/fs.h index 834c9e5113d9..d40e8e78bbd1 100644 --- a/include/linux/fs.h +++ b/include/linux/fs.h @@ -10,6 +10,7 @@ #include #include #include +#include #include #include #include @@ -768,6 +769,7 @@ struct file { */ union { struct list_head fu_list; + struct llist_node fu_llist; struct rcu_head fu_rcuhead; } f_u; struct path f_path; -- cgit v1.2.3-58-ga151 From fb4214db50b00558cc6e274c88b3f7325068e942 Mon Sep 17 00:00:00 2001 From: Oleg Nesterov Date: Mon, 8 Jul 2013 14:24:18 -0700 Subject: llist: fix/simplify llist_add() and llist_add_batch() 1. This is mostly theoretical, but llist_add*() need ACCESS_ONCE(). Otherwise it is not guaranteed that the first cmpxchg() uses the same value for old_entry and new_last->next. 2. These helpers cache the result of cmpxchg() and read the initial value of head->first before the main loop. I do not think this makes sense. In the likely case cmpxchg() succeeds, otherwise it doesn't hurt to reload head->first. I think it would be better to simplify the code and simply read ->first before cmpxchg(). Signed-off-by: Oleg Nesterov Cc: Al Viro Cc: Andrey Vagin Cc: "Eric W. Biederman" Cc: David Howells Cc: Huang Ying Cc: Peter Zijlstra Signed-off-by: Andrew Morton Signed-off-by: Al Viro --- include/linux/llist.h | 19 +++++++------------ lib/llist.c | 15 +++++---------- 2 files changed, 12 insertions(+), 22 deletions(-) (limited to 'include') diff --git a/include/linux/llist.h b/include/linux/llist.h index a5199f6d0e82..3e2b969d68f6 100644 --- a/include/linux/llist.h +++ b/include/linux/llist.h @@ -151,18 +151,13 @@ static inline struct llist_node *llist_next(struct llist_node *node) */ static inline bool llist_add(struct llist_node *new, struct llist_head *head) { - struct llist_node *entry, *old_entry; - - entry = head->first; - for (;;) { - old_entry = entry; - new->next = entry; - entry = cmpxchg(&head->first, old_entry, new); - if (entry == old_entry) - break; - } - - return old_entry == NULL; + struct llist_node *first; + + do { + new->next = first = ACCESS_ONCE(head->first); + } while (cmpxchg(&head->first, first, new) != first); + + return !first; } /** diff --git a/lib/llist.c b/lib/llist.c index 4a15115e90f8..4a70d120138c 100644 --- a/lib/llist.c +++ b/lib/llist.c @@ -39,18 +39,13 @@ bool llist_add_batch(struct llist_node *new_first, struct llist_node *new_last, struct llist_head *head) { - struct llist_node *entry, *old_entry; + struct llist_node *first; - entry = head->first; - for (;;) { - old_entry = entry; - new_last->next = entry; - entry = cmpxchg(&head->first, old_entry, new_first); - if (entry == old_entry) - break; - } + do { + new_last->next = first = ACCESS_ONCE(head->first); + } while (cmpxchg(&head->first, first, new_first) != first); - return old_entry == NULL; + return !first; } EXPORT_SYMBOL_GPL(llist_add_batch); -- cgit v1.2.3-58-ga151 From e9a17bd73a29e5323c37ec5ffe50fc0e825d3d03 Mon Sep 17 00:00:00 2001 From: Oleg Nesterov Date: Mon, 8 Jul 2013 14:24:19 -0700 Subject: llist: llist_add() can use llist_add_batch() llist_add(new, head) can simply use llist_add_batch(new, new, head), no need to duplicate the code. This obviously uninlines llist_add() and to me this is a win. But we can make llist_add_batch() inline if this is desirable, in this case gcc can notice that new_first == new_last if the caller is llist_add(). Signed-off-by: Oleg Nesterov Cc: Al Viro Cc: Andrey Vagin Cc: "Eric W. Biederman" Cc: David Howells Cc: Huang Ying Cc: Peter Zijlstra Signed-off-by: Andrew Morton Signed-off-by: Al Viro --- include/linux/llist.h | 14 ++++---------- 1 file changed, 4 insertions(+), 10 deletions(-) (limited to 'include') diff --git a/include/linux/llist.h b/include/linux/llist.h index 3e2b969d68f6..cdaa7f023899 100644 --- a/include/linux/llist.h +++ b/include/linux/llist.h @@ -142,6 +142,9 @@ static inline struct llist_node *llist_next(struct llist_node *node) return node->next; } +extern bool llist_add_batch(struct llist_node *new_first, + struct llist_node *new_last, + struct llist_head *head); /** * llist_add - add a new entry * @new: new entry to be added @@ -151,13 +154,7 @@ static inline struct llist_node *llist_next(struct llist_node *node) */ static inline bool llist_add(struct llist_node *new, struct llist_head *head) { - struct llist_node *first; - - do { - new->next = first = ACCESS_ONCE(head->first); - } while (cmpxchg(&head->first, first, new) != first); - - return !first; + return llist_add_batch(new, new, head); } /** @@ -173,9 +170,6 @@ static inline struct llist_node *llist_del_all(struct llist_head *head) return xchg(&head->first, NULL); } -extern bool llist_add_batch(struct llist_node *new_first, - struct llist_node *new_last, - struct llist_head *head); extern struct llist_node *llist_del_first(struct llist_head *head); #endif /* LLIST_H */ -- cgit v1.2.3-58-ga151 From a95e691f9c4a6e24fdeab6d7feae6d5411fe8a69 Mon Sep 17 00:00:00 2001 From: Al Viro Date: Sun, 14 Jul 2013 16:43:54 +0400 Subject: rpc_create_*_dir: don't bother with qstr just pass the name Signed-off-by: Al Viro --- include/linux/sunrpc/rpc_pipe_fs.h | 4 ++-- net/sunrpc/cache.c | 18 +++++------------- net/sunrpc/clnt.c | 20 ++++++++------------ net/sunrpc/rpc_pipe.c | 14 ++++++++------ 4 files changed, 23 insertions(+), 33 deletions(-) (limited to 'include') diff --git a/include/linux/sunrpc/rpc_pipe_fs.h b/include/linux/sunrpc/rpc_pipe_fs.h index a7b422b33eda..aa5b582cc471 100644 --- a/include/linux/sunrpc/rpc_pipe_fs.h +++ b/include/linux/sunrpc/rpc_pipe_fs.h @@ -73,12 +73,12 @@ extern ssize_t rpc_pipe_generic_upcall(struct file *, struct rpc_pipe_msg *, extern int rpc_queue_upcall(struct rpc_pipe *, struct rpc_pipe_msg *); struct rpc_clnt; -extern struct dentry *rpc_create_client_dir(struct dentry *, struct qstr *, struct rpc_clnt *); +extern struct dentry *rpc_create_client_dir(struct dentry *, const char *, struct rpc_clnt *); extern int rpc_remove_client_dir(struct dentry *); struct cache_detail; extern struct dentry *rpc_create_cache_dir(struct dentry *, - struct qstr *, + const char *, umode_t umode, struct cache_detail *); extern void rpc_remove_cache_dir(struct dentry *); diff --git a/net/sunrpc/cache.c b/net/sunrpc/cache.c index 80fe5c86efd1..b40f9567e628 100644 --- a/net/sunrpc/cache.c +++ b/net/sunrpc/cache.c @@ -1812,19 +1812,11 @@ int sunrpc_cache_register_pipefs(struct dentry *parent, const char *name, umode_t umode, struct cache_detail *cd) { - struct qstr q; - struct dentry *dir; - int ret = 0; - - q.name = name; - q.len = strlen(name); - q.hash = full_name_hash(q.name, q.len); - dir = rpc_create_cache_dir(parent, &q, umode, cd); - if (!IS_ERR(dir)) - cd->u.pipefs.dir = dir; - else - ret = PTR_ERR(dir); - return ret; + struct dentry *dir = rpc_create_cache_dir(parent, name, umode, cd); + if (IS_ERR(dir)) + return PTR_ERR(dir); + cd->u.pipefs.dir = dir; + return 0; } EXPORT_SYMBOL_GPL(sunrpc_cache_register_pipefs); diff --git a/net/sunrpc/clnt.c b/net/sunrpc/clnt.c index 5a750b9c3640..26456274b24e 100644 --- a/net/sunrpc/clnt.c +++ b/net/sunrpc/clnt.c @@ -128,9 +128,7 @@ static struct dentry *rpc_setup_pipedir_sb(struct super_block *sb, { static uint32_t clntid; char name[15]; - struct qstr q = { .name = name }; struct dentry *dir, *dentry; - int error; dir = rpc_d_lookup_sb(sb, dir_name); if (dir == NULL) { @@ -138,19 +136,17 @@ static struct dentry *rpc_setup_pipedir_sb(struct super_block *sb, return dir; } for (;;) { - q.len = snprintf(name, sizeof(name), "clnt%x", (unsigned int)clntid++); + snprintf(name, sizeof(name), "clnt%x", (unsigned int)clntid++); name[sizeof(name) - 1] = '\0'; - q.hash = full_name_hash(q.name, q.len); - dentry = rpc_create_client_dir(dir, &q, clnt); + dentry = rpc_create_client_dir(dir, name, clnt); if (!IS_ERR(dentry)) break; - error = PTR_ERR(dentry); - if (error != -EEXIST) { - printk(KERN_INFO "RPC: Couldn't create pipefs entry" - " %s/%s, error %d\n", - dir_name, name, error); - break; - } + if (dentry == ERR_PTR(-EEXIST)) + continue; + printk(KERN_INFO "RPC: Couldn't create pipefs entry" + " %s/%s, error %ld\n", + dir_name, name, PTR_ERR(dentry)); + break; } dput(dir); return dentry; diff --git a/net/sunrpc/rpc_pipe.c b/net/sunrpc/rpc_pipe.c index e7ce4b3eb0bd..63364cb5d11a 100644 --- a/net/sunrpc/rpc_pipe.c +++ b/net/sunrpc/rpc_pipe.c @@ -770,15 +770,17 @@ out_bad: } static struct dentry *rpc_mkdir_populate(struct dentry *parent, - struct qstr *name, umode_t mode, void *private, + const char *name, umode_t mode, void *private, int (*populate)(struct dentry *, void *), void *args_populate) { struct dentry *dentry; + struct qstr q = QSTR_INIT(name, strlen(name)); struct inode *dir = parent->d_inode; int error; + q.hash = full_name_hash(q.name, q.len); mutex_lock_nested(&dir->i_mutex, I_MUTEX_PARENT); - dentry = __rpc_lookup_create_exclusive(parent, name); + dentry = __rpc_lookup_create_exclusive(parent, &q); if (IS_ERR(dentry)) goto out; error = __rpc_mkdir(dir, dentry, mode, NULL, private); @@ -925,8 +927,8 @@ static void rpc_clntdir_depopulate(struct dentry *dentry) /** * rpc_create_client_dir - Create a new rpc_client directory in rpc_pipefs - * @dentry: dentry from the rpc_pipefs root to the new directory - * @name: &struct qstr for the name + * @dentry: the parent of new directory + * @name: the name of new directory * @rpc_client: rpc client to associate with this directory * * This creates a directory at the given @path associated with @@ -935,7 +937,7 @@ static void rpc_clntdir_depopulate(struct dentry *dentry) * later be created using rpc_mkpipe(). */ struct dentry *rpc_create_client_dir(struct dentry *dentry, - struct qstr *name, + const char *name, struct rpc_clnt *rpc_client) { return rpc_mkdir_populate(dentry, name, S_IRUGO | S_IXUGO, NULL, @@ -981,7 +983,7 @@ static void rpc_cachedir_depopulate(struct dentry *dentry) rpc_depopulate(dentry, cache_pipefs_files, 0, 3); } -struct dentry *rpc_create_cache_dir(struct dentry *parent, struct qstr *name, +struct dentry *rpc_create_cache_dir(struct dentry *parent, const char *name, umode_t umode, struct cache_detail *cd) { return rpc_mkdir_populate(parent, name, umode, NULL, -- cgit v1.2.3-58-ga151